Earlier quoted context omitted.
Aren't the IPs of the load balancers more or less fixed, though?
I'm not a netadmin, so don't know the innards of how it all works, but one day I watched a domain of ours pointing to an amazon load-balancer spread across multiple AZs - the IP address would change in DNS every so often, flipping between the AZs, presumably.
This is not hardware. This is software defined networking which you should apply one of the rules of good cloud design which is to expect failure - and this is a feature - IPs change. Deal with it. ;)
Further we have the TTL issue noted and an very interesting thing happens there. If you use a default of say 300 or more seconds for your FQDN in the DNS record which has even a CNAME to an ELB set of FQDNs (if multiple ELBs for example) then your going to have a condition at some times where that 300 seconds is still ticking down when the say 60 second TTL for the ELB's fqdn expires and or the ELB IP itself has changed. In that time span your resolved IP may be assigned to another ELB and traffic going to your platform hits some other platform. So perhaps at the first 1 second mark of your 300 second TTL start the AWS ELB TTL has expired and perhaps is assigned to nothing - your traffic gets failure to connect - then the IP is assigned to some other FQDN and your traffic hits some other ELB. (Your church patrons now get porn perhaps.) The flip side is true and interesting to watch in logs.
How to take advantage of this feature? Oh that is fun. Marketing? Route all such identifiable traffic to good bad or ugly ends?
This situation should be expected.